Point Cloud Neural Networks (PCNNs) have emerged as a vital tool for latency-sensitive 3D perception applications, such as autonomous driving and AR/VR. However, their inherent computational redundancy—arising from excessive global sampling/search operations and repeated feature updates/aggregations caused by shared neighbors—severely constrains execution efficiency. More critically, conventional redundancy elimination methods usually introduce operations that are highly GPU-unfriendly, resulting in high memory overhead, increased branch divergence, irregular memory access, and serial dependencies, which together pose a significant challenge to PCNN acceleration.

Unmanned aerial vehicles (UAVs), characterized by their agility, affordability, and flexible deployment, exhibit significant advantages in scenarios such as disaster monitoring, target tracking, and environmental data collection. However, the limited onboard energy of UAVs poses a major challenge for long-duration or large-scale missions. To address this issue, this paper proposes a dynamic trajectory planning framework for cooperative task search involving multiple UAVs. First, a UAV capability evaluation approach is developed to assess the relative performance of heterogeneous UAVs. Next, a density-aware clustering mechanism is employed to partition the search region based on spatial distance and regional density. After clustering, a sequential matching strategy is employed to assign UAVs with higher capabilities to larger or more complex task regions, ensuring efficient resource utilization. The problem is then formulated as a combinatorial optimization task, and a pointer network is designed to generate UAV trajectories. The network is trained using deep reinforcement learning to produce near-optimal paths, thereby minimizing the overall system cost. Experimental results confirm that the proposed method can substantially lower total task execution expenditure.

With the rapid development of the internet of things (IoT), IoT devices are now capable of real-time monitoring and collecting environmental and production data through integrated sensors. However, these devices often face challenges related to limited storage capabilities and transmission range. Furthermore, the widespread deployment of IoT devices has raised significant concerns regarding privacy security. To enhance data collection efficiency and ensure the security of location privacy, this study proposes a high altitude platform (HAP) data collection and trajectory design scheme that is aware of location privacy. Firstly, our scheme utilizes HAPs to quickly cover the collection area and transmit data in real time via satellites. Secondly, a differential privacy-based perturbation mechanism is applied to reduce the risk of location information leakage. Finally, the trajectory optimization problem, incorporating privacy awareness, is modeled as a Markov decision process (MDP) and solved using deep reinforcement learning (DRL) techniques to determine the movement decisions of the HAPs. Experimental results demonstrate that this scheme effectively protects location privacy while enhancing the efficiency and security of data collection.

As global demand for efficient network services increases, the limitations of traditional terrestrial wireless networks are becoming more apparent. Space-air-ground integrated networks (SAGIN) have emerged as a promising solution to advance next-generation network infrastructure. However, SAGIN faces significant security challenges—including the lack of a robust security architecture, trust and data reliability issues in multi-hop transmissions, and the need to enhance network performance without compromising security—rendering traditional boundary-based defenses inadequate. Therefore, we propose SECURELINK, a decentralized zero-trust architecture tailored for SAGIN, which replaces traditional perimeter defenses with a ”never trust, always verify” approach. This approach strengthens network security and flexibility through continuous verification and adherence to the principle of least privilege. SECURELINK integrates blockchain technology to establish a multi-layered security verification and data processing scheme, addressing the dynamic and decentralized features of SAGIN. Additionally, we introduce DFRIO, a zero-trust traffic offloading method based on decentralized federated reinforcement learning and blockchain, designed to enhance network performance within maintaining security. Simulation results demonstrate that our solution significantly enhances SAGIN’s defense capability without compromising network stability, outperforming the two baseline schemes by 18.3% and 42.6%, respectively.

Edge Learning environments, characterized by limited wireless resources, encounter significant bottlenecks in network performance, particularly in Federated Learning (FL) tasks. Current resource allocation strategies are primarily classified into “AI for Networking” and “Networking for AI”. However, both approaches fail to adequately address the interaction between network states and AI task requirements, thereby limiting their overall effectiveness. To address this, we propose a novel bidirectional dynamic collaborative optimization mechanism that enables real-time interaction between AI task performance and network states. This mechanism adjusts both AI task resource requirements and network configurations based on performance feedback, breaking away from traditional unidirectional optimization approaches. We introduce the AI-network unified algorithm, which incorporates data-driven dynamic sensing and enhances system adaptability and robustness, achieving self-optimization in edge learning. Theoretical analysis and simulation results demonstrate the significant advantages of our approach in simultaneously improving network resource utilization and AI task performance, providing an effective solution for the future wireless network.

Network device and system health management is the foundation of modern network operations and maintenance. Traditional health management methods, relying on expert identification or simple rule-based algorithms, struggle to cope with the heterogeneous networks (HNs) environment. Moreover, current state-of-the-art distributed fault diagnosis methods, which utilize specific machine learning techniques, lack multi-scale adaptivity for heterogeneous device information, resulting in unsatisfactory diagnostic accuracy for HNs. In this paper, we develop an LLM-assisted end-to-end intelligent network health management framework. The framework first proposes a multi-scale data scaling method based on unsupervised learning to address the multi-scale data problem in HNs. Secondly, we combine the semantic rule tree with the attention mechanism to propose a Multi-Scale Semanticized Anomaly Detection Model (MSADM) that generates network semantic information while detecting anomalies. Finally, we embed a chain-of-thought-based large-scale language model downstream to adaptively analyze the fault diagnosis results and create an analysis report containing detailed fault information and optimization strategies. We compare our scheme with other fault diagnosis models and demonstrate that it performs well on several metrics of network fault diagnosis.

Causal inference and root cause analysis play a crucial role in network performance evaluation and optimization by identifying critical parameters and explaining how the configuration parameters affect network key performance indicators (KPIs). Traditional performance evaluation methods can evaluate KPIs based on configuration parameters, but they are unable to explain how configuration parameters affect KPIs. Moreover, static causal discovery and inference methods are not directly applicable to dynamic networks. To address these challenges, we propose a self-iterative performance optimization method based on causal inference and interpretable network evaluation (CiiNet). CiiNet constructs causal graphs through change-point detection and hierarchical incremental causal discovery. Then, CiiNet introduces causal inference for critical parameter analysis (CPA). Using intervention analysis and regression-based parameter learning, CiiNet infers and evaluates the impact of critical parameters on KPIs. Based on the interpretable evaluation, CiiNet can further self-iteratively optimize the critical parameters for optimal network performance and dynamically obtain the optimal configurations. Our extensive experiments show that CiiNet outperforms other baseline methods regarding causal discovery, network performance evaluation, and CPA.

Fault diagnosis in microservice systems requires high availability, driving research towards multimodal learning that leverages heterogeneous monitoring data, including logs, metrics, and traces. These data inherently contain time series and data streams across various modalities. However, existing frameworks fail to exploit temporal dependencies in these dynamic streams effectively. First, existing methods rely too much on original time-domain metrics, which makes it difficult for them to capture periodic patterns and sudden events. Second, most language-bound integrations treat timestamps only as sequential labels or numerical inputs, which ignores the rich contextual information within timestamps. As a result, existing models struggle to comprehend the temporal sequences and data characteristics associated with faults in multimodal data. In this work, we introduce FD-TE Diagnosis, a framework for microservice fault diagnosis that addresses these challenges by applying frequency domain feature analysis and time encoding. To enhance the detection of abnormal events, we integrate a frequency-domain approach that utilizes the Fast Fourier Transform (FFT) to extract robust features from metric data. Next, we encode the timestamps of all events using a dedicated time encoding layer. These temporal representations are then incorporated to strengthen event embedding for fault inference. Experimental results demonstrate that our method enhances the sensitivity of multimodal models in interpreting temporal features, resulting in improved diagnostic outcomes.

We propose using the Poisson point process (PPP) for accurate modeling in Space-air-ground integrated networks (SAGIN). SAGIN is experiencing significant growth, aiming for ultra-fast speeds, low latency, and integration of AI, IoT, and other emerging technologies. To achieve better performance than existing hybrid bands in SAGIN, we utilize three distinct frequency bands—Radio frequency (RF), Terahertz (THz), and Free-space optical communication (FSO)—to compensate for their shortcomings effectively. Then, we derive not only the coverage probability but also the rate coverage probability formula of the network and utilize Newton's method to deduce the most optimal channel allocation scheme. Moreover, we use comparison experiments between different frequency bands, altitudes, and other conditions to affirm its reliability and effectiveness in advancing SAGIN coverage and rate coverage probability. The results show that our proposed scheme performs up to 3 times better than a single-spectrum scheme and 2 times better than the existing mixed method.

The integration of Time-Sensitive Networking (TSN) with 5G technology provides Industrial IoT (IIoT) systems with essential low latency, high flexibility, and reliability. However, a key challenge in combining 5G and TSN is the deterministic scheduling of cross-domain flows, which requires precise time synchronisation and the ability to handle unpredictable changes in wireless channels. To address this challenge, we propose an online asynchronous scheduling mechanism. This mechanism is implemented at the 5G-TSN gateway, dynamically allocating TSN network time slot resources to enhance the network's deterministic scheduling capability in the presence of time asynchrony and network fluctuations. Extensive simulations on the OMNeT++ platform demonstrate that our online asynchronous algorithm effectively utilises network resources, reduces delays caused by wireless fluctuations and time asynchrony, and improves network throughput.

The rapid growth of graph-structured data necessitates partitioning and distributed storage across decentralized systems, driving the emergence of federated graph learning to collaboratively train Graph Neural Networks (GNNs) without compromising privacy. However, current methods exhibit limited performance when handling hypergraphs, which inherently represent complex high-order relationships beyond pairwise connections. Partitioning hypergraph structures across federated subsystems amplifies structural complexity, hindering high-order information mining and compromising local information integrity. To bridge the gap between hypergraph learning and federated systems, we develop FedHGL, a first-of-its-kind framework for federated hypergraph learning on disjoint and privacy-constrained hypergraph partitions. Beyond collaboratively training a comprehensive hypergraph neural network across multiple clients, FedHGL introduces a pre-propagation hyperedge completion mechanism to preserve high-order structural integrity within each client. This procedure leverages the federated central server to perform cross-client hypergraph convolution without exposing internal topological information, effectively mitigating the high-order information loss induced by subgraph partitioning. Furthermore, by incorporating two kinds of local differential privacy (LDP) mechanisms, we provide formal privacy guarantees for this process, ensuring that sensitive node features remain protected against inference attacks from potentially malicious servers or clients. Experimental results on seven real-world datasets confirm the effectiveness of our approach and demonstrate its performance advantages over traditional federated graph learning methods.

The rapid proliferation of industrial equipment and its widespread deployment across diverse sectors have introduced substantial challenges for fault diagnosis. Current equipment operates under a range of disparate fault coding standards, marked by pronounced heterogeneity and fragmentation—particularly in Identification and classification of equipment and faults. The lack of a standardized representation has been shown to impede cross-domain data integration and to constrain the adaptability and generalizability of existing diagnostic models in complex, multi-source environments. To address these limitations, this study proposes FUSE74 — a novel Fault Unification and Semantic Encoding scheme that standardizes fault information using a structured 74-bit representation. This scheme defines a generalized and extensible coding structure, supported by a rule-based mapping mechanism that links fault codes to semantic representations. Such a design enables the standardized expression of fault-related information across heterogeneous systems. Building upon this foundation, the paper further introduces a diagnostic framework driven by a large language model (LLM), which utilizes the LLM’s semantic reasoning capabilities to perform automated fault analysis, health assessment, and maintenance recommendation. Experimental evaluations demonstrate the proposed framework’s effectiveness in achieving robust cross-equipment adaptability and high diagnostic accuracy, thereby providing a practical solution for intelligent fault management in complex industrial contexts.

Anomaly detection in multivariate time series data is critical across a variety of real-life applications. The predominant anomaly detection techniques currently rely on reconstruction-based methods. However, these methods often overfit the abnormal pattern and fail to diagnose the anomaly. Although some studies have attempted to prevent the incorrect fitting of anomalous data by enabling models to learn the trend of data variations, they fail to account for the dynamic nature of data distribution. This oversight can lead to the erroneous reconstruction of anomalies that do not exist. To address these challenges, we propose RTdetector, a Transformer-based time series anomaly detection model leveraging reconstruction trends. RTdetector employs a novel global attention mechanism based on reconstruction trends to learn distinguishable attention from the original sequence, thereby preserving the global trend information intrinsic to the time series. Additionally, it incorporates a self-conditioning transformer, based on reconstruction trend enhancement to achieve superior predictive performance. Extensive experiments on four datasets demonstrate that RTdetector achieves state-of-the-art results in multivariate time series data anomaly detection. Our code is available at https://github.com/CSUFUNLAB/RTdetector.

In satellite Internet of Things (IoT), the remote sensing satellites capture images and then transmit them to a ground station through low Earth orbit (LEO) communication satellites for model inference. However, this process results in significant transmission latency and communication overhead. In response, researchers have proposed various satellite on-orbit model inference methods. Nonetheless, the limited computation capacity and memory space of a single remote sensing satellite impose processing delays when dealing with large quantities of high-resolution images, thereby making it difficult to ensure real-time service. To tackle this issue, we propose an on-orbit deep neural network (DNN) distributed inference framework for remote sensing images in satellite IoT, leveraging the availability of numerous LEO computing satellites. Designing such a framework involves two crucial questions: first, determining which LEO satellites should participate in distributed DNN inference, and second, how to partition the images among the selected LEO satellites. To address these questions, we formulate the distributed inference process as a mixed integer nonlinear optimization problem, which is known to be NP-hard. The objective is to minimize overall energy consumption while ensuring that the distributed inference is accomplished when the satellite dynamic network remains unchanged. We initially propose a dynamic optimization algorithm that derives the optimal solution with rigorous theoretical guarantees. Subsequently, to reduce computational complexity, we introduce an approximate solution based on an improved simulated annealing algorithm. We demonstrate that the approximate algorithm performs within a limited range of the optimal algorithm. Finally, we build a heterogeneous testbed based on Kubernetes and conduct extensive experiments to validate that our proposed algorithms reduce energy consumption by an average of 24.63% and 25.98% on the Faster-RCNN inference model, 47.09% and 47.51% on the RetinaNet inference model, and 53.36% and 48.08% on the Yolov5 inference model on the two datasets compared to the baselines.

Vehicle trajectory prediction plays a crucial role in IoT-based intelligent transportation systems, which can effectively address key issues, such as driving safety and multivehicle collaboration. However, the sensitivity of trajectory data and the reluctance of data holders to share it constrain the prediction model’s ability to capture vehicle behavior patterns in different scenarios. To address the above problems, we propose a blockchain-enabled asynchronous federated proximal policy optimization framework (BE-AFPPO) for the trajectory prediction of self-driving vehicles. First, we propose a curiosity proximal policy optimization (C-PPO) algorithm. The method utilizes a driven exploration strategy to actively motivate the intelligent agent to explore the unknown state space. The avoidance policy model reaches a local optimum when processing trajectory data. In addition, we design historical gated recurrent unit (GRU) and future GRU as input layers. The target’s historical motion features and future trajectory features are extracted, respectively. Then, various data is received through asynchronous federated learning. This model can fully learn the vehicle’s behavior patterns in different scenarios, which improves prediction accuracy. Based on this, we develop a blockchain-based dynamic group practical Byzantine fault tolerance (DG-PBFT) consensus algorithm. This enhances the credibility and integrity of the data while enriching the sources of trajectory data. Finally, we perform the experiments on the publicly available dataset nuScenes. The results demonstrate that the proposed method improves the robustness and accuracy of trajectory prediction.

Point cloud neural networks are gaining increasing attention in emerging 3-D computer vision applications, such as autonomous driving, robotics, and virtual reality. Many customized accelerators for 3-D point clouds have been developed to pursue superior time and energy efficiencies. In this work, we reveal that spatially adjacent points in a 3-D point cloud show similar feature values and relationships, implying substantial redundant computations and memory accesses, while which have been previously ignored. To reduce such redundancies, we propose SimDiff, an algorithm-accelerator co-design framework that boosts 3-D point cloud processing by cleverly leveraging spatial similarity toward excellent speedup and energy efficiency. On the algorithm side, we design a novel similarity-aware differential point cloud neural network (dubbed SD-PCNet). Differing from the standard flow of mainstream point cloud networks, it abstracts a brand-new execution flow for point cloud processing by utilizing spatial similarity among points and dynamic differential execution. On the accelerator side, we propose SD-PCAcc, a supporting accelerator to convert algorithm-level redundancy reductions into performance enhancements. On the deployment side, we propose efficient strategies for network-to-accelerator mapping and scheduling, high-bandwidth memory (HBM) channel allocation, and core component reconfiguration, facilitating the proposed methodologies into practical implementation. Extensive evaluation results show that, with preserved accuracy, our SimDiff gains an average of$3.2\times $speedup and$3.1\times $energy efficiency compared to the state-of-the-art competitors.

Existing anti-jamming strategies for unmanned aerial vehicle (UAV) networks largely assume homogeneity among UAVs, neglecting the differences in hardware configurations, task requirements, and environmental adaptability. In the face of such heterogeneity, these strategies often fail to effectively counter intelligent jamming and co-channel interference. To address this issue, this paper proposes an intelligent anti-jamming framework designed specifically for the heterogeneous UAV network, allowing each UAV to autonomously adjust its transmission channel and power based on its hardware capabilities and task requirements in a distributed environment. This aims to optimize communication efficiency and reduce energy consumption. We formulate the anti-jamming problem as an adversarial game and confirm the existence of a unique equilibrium point within this model. Moreover, we introduce the novel Personalized Federated Soft Actor-Critic (PFSAC) algorithm, which combines the global model with local models to customize personalized anti-jamming strategies for each UAV, significantly enhancing network performance in complex jamming environments. Simulation results indicate that compared to other methods, our proposed algorithm significantly enhances the anti-jamming capability of heterogeneous UAV networks and performs better than them.

Highly dynamic heterogeneous networks (HDHNs), characterized by high node mobility and heterogeneity, frequently experience complex and recurrent network faults. Conventional centralized fault diagnosis methods demand real-time collection of extensive network-wide data, while distributed approaches often exhibit limited fault detection capabilities. Additionally, machine learning-based fault diagnosis methods are challenged by the scarcity of labeled fault samples required for training. To address these limitations, this study proposes a semi-distributed network fault diagnosis architecture based on a digital twin network (DTN). The proposed architecture facilitates the extraction of a comprehensive labeled fault dataset that closely replicates real-world network conditions. Using this dataset, we perform centralized training of an enhanced anomaly detection model, FTS-LSTM, to infer fault types at the node level. To overcome the drawbacks of both centralized and distributed approaches, we further introduce a semi-distributed fault diagnosis algorithm (SDFD) that integrates fault types and severity levels identified by nodes to infer overall network faults. The proposed fault diagnosis scheme is validated on a semi-physical DTN simulation platform, demonstrating its effectiveness in realistic scenarios.

Service function chain placement is crucial to support services flexibility and diversity for different users and vendors. Specifically, this problem is proved to be NP-hard. Existing deep reinforcement learning based methods either can only handle a limited number of objectives, or their training time are too long. Concomitantly, they are unable to satisfy when the number of objectives is dynamic. It is necessary to model service function chain placement as a multi-objective problem. The multi-objective problem can decomposed into multiple sub-problems by the weight vectors. In this paper, we first reveal the relationship between weight vectors and solution position, which can reduce the training time to gain a better placement model. Then, we design a novel algorithm for the service function chain placement problem, called rzMODRL. The weight vectors are divided into zones for training in parallel, and the order is defined for the final models located at the end of a training process, which can save time and improve the quality of the model. Dynamic objective placement method is based on the high-dimensional model to avoid retraining for a low-dimensional placement. Evaluation results show that the proposed algorithms improve the service acceptance ratio up to 32% and the hyper-volume values with 14% in the multi-objective service function chain placement, where hyper-volume has been widely applied to evaluate the convergence and diversity simultaneously in multi-objective optimization. And it is also effective in solving the dynamic objective service function chain placement problem that the difference of average hyper-volume values is 10.44%.

With the advancements in six-generation (6G) communication technology, there is a growing need for comprehensive and interpretable network performance evaluation for network optimization. Traditional evaluation methods often overlook uncertainties and are limited to a single time scale or performance dimension, while the recent machine learning-based method lacks interpretability. To address this issue, we propose a multidimensional performance evaluator for interpretable and traceable network performance evaluation (MPITE). MPITE, constructed with a three-layer evaluation model incorporating physical, logical, and causal topology structures, reflects the causal relationship of communication system configurations, the changing network states, and performance metrics. We introduce a multidimensional performance index that considers value, time, and certainty dimensions to evaluate network performance comprehensively. We propose interpretable Bayesian theory-based network inference algorithms to derive network certainty for interpretable network performance evaluation. Then, we intelligently derive optimal network configuration parameters through reverse inferencing for network tracing. Experimental results demonstrate the advantage, interpretability, and traceability of MPITE.

The Space-Air-Ground integrated network (SAGIN) has gained significant attention due to the explosive growth in mobile data traffic. In this network, Unmanned Aerial Vehicles (UAVs) play a critical role as air relay nodes, bridging ground and space networks. However, challenges arise from the dynamic position angles between UAVs and satellites, as well as fixed UAV altitudes, limiting air-to-space transmission capacity. Moreover, the finite UAV battery capacity carries the risk of energy interruptions during SAGIN transmissions. To address these issues, we propose an integrated model that considers UAV channel fading, energy consumption, and harvesting. This model allows us to comprehensively analyze SAGIN transmission performance. Within this framework, we calculate the UAV energy outage probability and signal-to-noise ratio (SNR) outage probability for SAGIN uplink transmission. Based on our network performance analysis, we derive an expression for the optimal UAV altitude, ensuring uninterrupted energy supply and preventing SNR outage. To assess the fairness of SAGIN transmission performance, we compare the capabilities of Ground-to-Air-to-Space and Ground-to-Space transmissions. Additionally, we provide closed-form expressions for the transmission time gap in both scenarios. Our numerical results validate the accuracy of these derived expressions and evaluate how key parameters impact the optimal UAV altitude in the SAGIN uplink.

In this article, we propose a downlink communication scheme for large-scale high-interference unmanned aerial vehicle (UAV) swarm network based on nonorthogonal multiple access (NOMA), clustering, and reinforcement learning (RL). Since a large number of UAVs increases the complexity of downlink communication, we first introduce a load-balancing fuzzy C-Means (LB-FCMs) algorithm for UAV clustering. Downlink communication consists of three stages: 1) UAV clustering; 2) data aggregation; and 3) data offloading. We have two goals: 1) maximize the data aggregation rate of the network while ensuring fairness of UAVs’ spectrum access for UAV-to-UAV (U2U) communications during data aggregation and 2) maximize network data offloading rate while ensuring ground station priority for UAV-to-ground (U2G) communications during data offloading. To address these two problems, first, we introduce uplink NOMA and downlink NOMA to eliminate part of the intrasystem interference, respectively. Then, we propose a multiagent RL framework for optimizing channel, transmit power, and trajectory scheduling (MARL-CPT). MARL-CPT consists of two parts of the algorithm, which solve the optimization problems in two stages, respectively. Simulation results show that our proposed method outperforms random decision-making and polling-based single-agent RL methods in terms of final score, fairness, and priority. For trajectory scheduling during data offloading, our method finds the optimal hover position while taking less than half the time compared to single-agent RL methods.

With the continuous development of airborne communication, the demand for efficient internet access on airplanes has been increasing. To enhance the communication service quality for airborne users and address the challenge of high content request latency, a three-layer communication structure with satellite and terrestrial-assisted caching is proposed. In this structure, satellites, base stations, and aircraft cooperatively cache content to serve users aboard airplanes. Considering variations in request preferences, content popularity in aircraft, base stations, and satellites, as well as constraints related to cache space and communication duration, a content placement problem is formulated to minimize the total system latency. To tackle this problem, the content placement and delivery process is modeled as a Markov decision process (MDP). Subsequently, a Deep Reinforcement Learning (DRL)-based airborne communication cache placement algorithm named ACCP is introduced to derive optimal content placement decisions. Additionally, we expedite the convergence of ACCP with a prioritized experience replay mechanism and reduce time complexity using a sumTree data structure. Simulation results demonstrate that the proposed method significantly improves cache hit rate and reduces content delivery latency compared to other schemes.

In the domain of medical diagnostics, precise identification of various skin and oral diseases is vital for effective patient care. In particular, Mpox is a potentially dangerous viral disease with zoonotic origins, capable of human-to-human transmission, underscoring the urgency of precise diagnostic methods for timely intervention. This paper introduces a novel approach named the Choquet Fuzzy Integral-based Ensemble (CFI-Net) for accurate classification of skin diseases, with a specific emphasis on detecting Mpox, foot ulcers, and various mouth and oral diseases. Our methodology begins with Transfer Learning, enhancing the classification capabilities of base classifiers (DenseNet169, MobileNetV1 and DenseNet201) by incorporating additional layers. Subsequently, we aggregate the prediction scores from each base classifier using the Choquet fuzzy integral (CFI) to derive the final predicted labels, thus ensuring dynamic and robust predictions. Fuzzy measures, a crucial component of this fuzzy integral-based ensemble method, are typically determined through manual experimentation in previous approaches. However, in our study, we have tackled the challenge of manual tuning by employing meta-heuristic optimization algorithm to precisely configure the fuzzy measures for optimal performance. A rigorous evaluation is conducted on four publicly available datasets, encompassing two Mpox datasets, a foot ulcer dataset, and a mouth and oral disease dataset. The experiments reveal the remarkable effectiveness of CFI-Net in significantly improving disease classification accuracy. Additionally, we employ Grad-CAM analysis to provide insights into the decision-making processes of our models. Our findings underscore the exceptional performance of CFI-Net, achieving accuracy rates of 98.06% and 94.81% for Mpox detection, 99.06% for foot ulcer detection, and an impressive 99.61% for mouth and oral disease classification. This research not only contributes to the advancement of disease diagnosis but also demonstrates the effectiveness of ensemble learning techniques coupled with fuzzy integral-based fusion in enhancing diagnostic accuracy.

The Space-Air-Ground Integrated Network (SAGIN) plays a pivotal role as a comprehensive foundational network communication infrastructure, presenting opportunities for highly efficient global data transmission. Nonetheless, given SAGIN's unique characteristics as a dynamically heterogeneous network, conventional network optimization methodologies encounter challenges in satisfying the stringent requirements for network latency and stability inherent to data transmission within this network environment. Therefore, this paper proposes the use of differentiated federated reinforcement learning (DFRL) to solve the traffic offloading problem in SAGIN, i.e., using multiple agents to generate differentiated traffic offloading policies. Considering the differentiated characteristics of each region of SAGIN, DFRL models the traffic offloading policy optimization process as the process of solving the Decentralized Partially Observable Markov Decision Process (DEC-POMDP) problem. The paper proposes a novel Differentiated Federated Soft Actor-Critic (DFSAC) algorithm to solve the problem. The DFSAC algorithm takes the network packet delay as the joint reward value and introduces the global trend model as the joint target action-value function of each agent to guide the update of each agent's policy. The simulation results demonstrate that the traffic offloading policy based on the DFSAC algorithm achieves better performance in terms of network throughput, packet loss rate, and packet delay compared to the traditional federated reinforcement learning approach and other baseline approaches.

Space-air-ground integrated networks (SAGIN) incorporating multi-band terahertz (THz) and radio frequency (RF) communication have gained increasing attention in the 6G era. However, the heterogeneity, self-organization, and time-variability of SAGIN pose challenges in accurately modeling, quantitatively analyzing, and optimizing these networks. Additionally, the dynamic topology and randomness of the nodes, including low-earth orbit satellites and high-altitude platforms, make the conventional THz/RF channel allocation method of terrestrial networks unsuitable for SAGIN. To address these challenges, we construct an accurate model of SAGIN based on the binomial point process (BPP) model in stochastic geometry. Subsequently, we analyze the network performance, specifically the joint coverage and transmission rate, through the proposed model. We then propose a simulated annealing algorithm-based optimization algorithm to achieve the optimal THz and RF channel allocation, effectively improving the joint coverage and transmission rate performance. Our simulation results demonstrate the effectiveness of the optimization algorithm and provide insights into the deployment rules of SAGIN.

Space-air-ground integrated network (SAGIN) is a promising network architecture for next-generation wireless networks, which combines satellite networks, aerial networks, and terrestrial networks to enable ubiquitous global network services to ground users and improve connectivity for wide deployment wireless applications. Also, free-space optical (FSO) communication with the advantages of low deployment cost, energy efficiency, and extremely high-speed data-delivering capability has attracted more attention recently. However, data transmission efficiency in SAGIN is still limited by the dynamic time-varying network topology and data transmission link connection. In this paper, we construct an FSO/radio frequency (RF) space-air-ground integrated network to enable large-scale and high-speed data transmission as well as degrade the burden of terrestrial networks. In addition, a deep-Q network-based reinforcement learning with an experience replay memory mechanism is proposed to execute dynamic hybrid routing by evaluated rewards. The simulation results show that the proposal achieves significant network performance compared with baseline methods.

Massive videos are released every day particularly through video-focused social media apps such as TikTok. This trend has fostered the quick emergence of video retrieval systems, which provide cloud-based services to retrieve similar videos using machine learning techniques. Adversarial example (AE) attacks have been shown to be effective on such systems by perturbing an unaltered video subtly to induce false retrieval results. Such AE attacks can be easily detected because the adversarial perturbations are all over pixels and frames. In this paper, we propose DUO, a stealthy targeted black-box AE attack which uses DUal search Over frame-pixel to generate sparse perturbations and improve stealthiness. DUO is motivated by two observations: only “key frames” in a video decide model predictions, and different pixels and frames contribute far differently to AEs. We implement DUO into a sequential attack pipeline consisting of two components (i.e., SparseTransfer and SparseQuery) built upon such intuitions. In particular, DUO uses SparseTransfer to generate initial perturbations and then SparseQuery to further rectify them. Extensive evaluations on two popular datasets confirm the higher efficacy and stealthiness of DUO over existing AE attacks on video retrieval systems. In particular, we show that DUO achieves higher precision while significantly reducing adversarial perturbations by more than ×100 than the state-of-the-art AE attack.

Sentiment analysis is a crucial task that aims to understand people's emotional states and predict emotional categories based on multimodal information. It consists of several subtasks, such as emotion recognition in conversation (ERC), aspect-based sentiment analysis (ABSA), and multimodal sentiment analysis (MSA). However, unifying all subtasks in sentiment analysis presents numerous challenges, including modality alignment, unified input/output forms, and dataset bias. To address these challenges, we propose a Task-Specific Prompt method to jointly model subtasks and introduce a multimodal generative framework called UniSA. Additionally, we organize the benchmark datasets of main subtasks into a new Sentiment Analysis Evaluation benchmark, SAEval. We design novel pre-training tasks and training methods to enable the model to learn generic sentiment knowledge among subtasks to improve the model's multimodal sentiment perception ability. Our experimental results show that UniSA performs comparably to the state-of-the-art on all subtasks and generalizes well to various subtasks in sentiment analysis.

With the explosive growth of mobile devices and communication facilities, electromagnetic interference (EMI) has become a common phenomenon affecting the communication band. Based on the shielding capability of electromagnetic bands in EMI, frequency selective surfaces (FSSs) are used to shield or suppress specific electromagnetic bands. Additionally, EMI can be negative control and may change the EMI band. Thus, a single FSS cannot effectively shield EMI due to its limited shielding capacity. To address this issue, we first construct a novel interference shielding model to guard the target area. The related shielding problem is modeled as the UAV-driven FSS (UFSS) configuration problem. Second, we propose an intelligent configuration method based on a stochastic game to solve the configuration optimization problem effectively. In the proposed method, we model the interaction between UFSSs and interferers as a stochastic game, where we provide each UFSS with two different options for updating its shielding configuration strategy. According to the shielding configuration strategy generated by the proposed stochastic game, we propose a square loop resource allocation model based on resource constraints to promote each UFSS to update its square loop. Finally, the numerical results and analysis show that our proposed method is more effective and feasible than other band shielding configuration schemes.

In the context of the 5G/6G mobile network, high levels of requirements such as ultra-high data transmission rate, support for the high mobility node and seamless connection need to be handled. Additionally, ensuring user quality of service (QoS) in high-density and high-traffic mobile networks presents a significant challenge. Unmanned aerial vehicles (UAVs) have emerged as key components in providing flexible assistance in aerial spaces. To further enhance the network performance in dynamic and heterogeneous environments, an intelligent resource allocation strategy with low communication overhead is essential. In this paper, we construct a UAV-assisted mobile network to provide efficient communication for all mobile users in high-density and high-traffic environments, at the same time, a digital twin-empowered dynamic resource allocation strategy based on online training with low communication overhead is proposed. Our proposal employs digital twin-empowered multi-task learning to meet various resource allocation requirements for different node types. Moreover, we propose a deep-Q network-based reinforcement learning mechanism with experience replay memory to execute resource allocation decisions based on evaluated rewards. The simulation results show that the proposal achieves significant network performance compared with baseline algorithms.

The sixth generation (6G) is conceived to address the expected high level of requirements (such as ultra-high-data-transmission rate, support for the highest moving speed and seamless connection, etc.) in the next decade and beyond. In the context of 6G, a large number of Industrial Internet of Things (IoT) (IIoT) devices may access the network, and thanks to the rapid development of artificial intelligence make smart manufacturing has the opportunity to be realized. However, a large number of IoT devices, the tremendous volume of data, the heterogeneous nature of devices, and the increasing concerns of privacy challenge the efficient management and quality of services in IIoT. To address these problems, in this article, a device-to-device (D2D) communication-aided digital twin edge network is proposed, where edge computing is introduced to bring computing and storage resources near to the end devices, and digital twin is utilized to fill the gap between physical and virtual space and D2D communication is applied to assist resource limited IoT devices to achieve normal communication. Moreover, digital twin-empowered federated reinforcement learning is leveraged to provide privacy awareness and decentralized resource allocation strategy training on D2D communication links to further improve network performance. The simulation results show that the proposal achieves significant network performance compared with baseline algorithms.

In the face of intensive computing tasks with massive data, cloud computing is difficult to provide high-quality services. Edge computing extends cloud services to the edge of the network by introducing edge devices between terminal devices and the cloud. For limited edge server resources, it is especially important to optimize offload strategies by accurately predicting the load on the terminal device. This paper proposes a C-LSTM prediction model based on deep neural network to predict the CPU utilization of terminal equipment in the future, and then proposes a distributed greedy algorithm for offloading decision. The simulation results show that the accuracy of C-LSTM prediction model is higher than other baseline models, reduces energy consumption and delay, and provides high-quality computing services.

Single image super-resolution (SISR) is a challenging ill-posed problem due to multiple high-resolution (HR) images can degenerate into the same low-resolution (LR) image. However, existing deep learning-based super-resolution (SR) methods always have blurred edge structures in the restored images. In addition, they mainly build more profound and more complex convolutional neural networks (CNN), which leads to substantial computational overhead. To address these issues, we propose the feature fusion super-resolution network (FFSRN) that uses the gradient map of the image to guide the restoration. In FFSRN, we propose the split and shuffle concat block (SSCB), which can extract rich features while controlling the model size and computational effort. We also introduce gradient branching to provide additional structural priors for the reconstruction process to restore high-resolution gradient mapping. Experimental results show that this method has a better peak signal-to-noise ratio, computational overhead and visual quality than the existing super-resolution algorithms. Code is available at https://github.com/Qyzs/FFSRN.

Space-Air-Ground Integrated Networks (SAGIN) is considered as the key structure of the next generation network. The space satellites and air nodes are the potential candidates to assist and offload the terrain transmissions. However, due to the high mobility of space and air nodes as well as the high dynamic of network traffic, the conventional traffic offloading strategy is not applicable for the high dynamic SAGIN. In this paper, we propose a reinforcement learning based traffic offloading for SAGIN by considering the high mobility of nodes as well as frequent changing network traffic and link state. In the proposal, a double Q-learning algorithm with improved delay-sensitive replay memory algorithm (DSRPM) is proposed to train the node to decide offloading strategy based on the local and neighboring historical information. Furthermore, a joint information collection with hello package and offline training mechanism is proposed to assist the proposed offloading algorithm. The simulation shows that the proposal outperforms conventional offloading algorithms in terms of signaling overhead, dynamic adaptivity, packet drop rate and transmission delay.

In the future era of intelligent networks, communication technology and network architecture need to be further developed to provide users with high-quality services. The Space-Air-Ground Integrated Networks (SAGIN) is seen as a potential architecture to provide ubiquitous communication and drive the era of the intelligent global network. The space and air segments in SAGIN can assist in offloading traffic from the ground segment. However, in a highly dynamic and heterogeneous network like SAGIN, offloading decisions are easily affected by the incorporated/malicious nodes. How to ensure security and improve network performance becomes a critical problem. In this paper, we address the above problem by jointly using blockchain and federated reinforcement learning (FRL). Firstly, we propose a blockchain-based secure federated learning framework that combines topology information chain and model chain to assist traffic offloading. Then, we propose a node security evaluation and an enhanced practical byzantine fault tolerance (EPBFT) algorithm to secure the traffic offloading process. Furthermore, we describe the traffic offloading problem as a Markov decision problem (MDP) and employ the Blockchain-based Federated Asynchronous Advantage Actor-Critic (BFA3C) algorithm to solve this problem. Finally, the simulation results show that the BFA3C-based algorithm used in SAGIN with/without malicious nodes achieves superior performance in terms of latency and security.

Recently, the 5G is widely deployed for supporting communications of high mobility nodes including train, vehicular and unmanned aerial vehicles (UAVs) largely emerged as the main components for constructing the wireless heterogeneous network (HetNet). To further improve the radio utilization, the Time Division Duplex (TDD) is considered to be the potential full-duplex communication technology in the high mobility 5G network. However, the high mobility of users leads to the high dynamic network traffic and unpredicted link state change. A new method to predict the dynamic traffic and channel condition and schedule the TDD configuration in real-time is essential for the high mobility environment. In this paper, we investigate the channel model in the high mobility and heterogeneous network and proposed a novel deep reinforcement learning based intelligent TDD configuration algorithm to dynamically allocate radio resources in an online manner. In the proposal, the deep neural network is employed to extract the features of the complex network information, and the dynamic Q-value iteration based reinforcement learning with experience replay memory mechanism is proposed to adaptively change TDD Up/Down-link ratio by evaluated rewards. The simulation results show that the proposal achieves significant network performance improvement in terms of both network throughput and packet loss rate, comparing with conventional TDD resource allocation algorithms.

As a powerful tool, the vehicular network has been built to connect human communication and transportation around the world for many years to come. However, with the rapid growth of vehicles, the vehicular network becomes heterogeneous, dynamic, and large scaled, which makes it difficult to meet the strict requirements, such as ultralow latency, high reliability, high security, and massive connections of the next-generation (6G) network. Recently, machine learning (ML) has emerged as a powerful artificial intelligence (AI) technique to make both the vehicle and wireless communication highly efficient and adaptable. Naturally, employing ML into vehicular communication and network becomes a hot topic and is being widely studied in both academia and industry, paving the way for the future intelligentization in 6G vehicular networks. In this article, we provide a survey on various ML techniques applied to communication, networking, and security parts in vehicular networks and envision the ways of enabling AI toward a future 6G vehicular network, including the evolution of intelligent radio (IR), network intelligentization, and self-learning with proactive exploration.

Deep learning has emerged as a popular Artificial Intelligence (AI) technique to make conventional cyber physical systems become intelligent and sustainable. Recently, deep learning has been widely used in the network domain. With the aid of powerful deep neural networks, the communication network can carry out packets forwarding actions intelligently to avoid possible failure and congestion. However, with the high computing cost and process limitation in only the static network scenario, the existing deep learning based network traffic control algorithms cannot satisfy the sustainable requirement of next generation large scale dynamic network. To conquer the existing problems, a novel spatial-temporal value network aided deep learning based intelligent traffic control algorithm referred as ST-DeLTA is proposed in this paper. In ST-DeLTA, the value matrix and spatial temporal training model (ST model) are employed to intelligently extract the spatial as well as temporal features of traffic patterns and make adaptive packets forwarding decision in large scale and dynamic networks. The mathematical analysis gives the computing cost reduction of our proposal, and the computer simulation demonstrates that our proposal has significantly better training and network performance compared with traditional algorithms in terms of training accuracy, transmission throughput, and average packets loss rate.

Disasters, such as earthquakes, typhoons, and tsunamis, usually cause extreme damages to the communication infrastructures, which results in a heavy recovery workload and seriously affects people's life. The disaster recovery networks play a critical role to reduce the loss caused by the disasters. However, the suddenly varying traffic demand and limited resources after disasters may lead to the repetitive reconfigurations for running the existing packet forwarding strategies, such as the shortest path algorithms. To handle this problem, it is necessary to adopt the deep learning technique to develop a disaster-resilient solution. In this paper, we utilize the deep reinforcement learning technique to propose a self-adaptive routing method for the Movable and Deployable Resource Unit (MDRU) based backbone network. Compared with existing deep learning based routing strategy, our proposal can adapt to the sudden network errors. Moreover, we also analyze the deployment manner and consider a centralized control structure to significantly balance the traffic.

Recently, the rapid advancement of high computing platforms has accelerated the development and applications of artificial intelligence techniques. Deep learning, which has been regarded as the next paradigm to revolutionize users' experiences, has attracted networking researchers' interests to relieve the burden due to the exponentially growing traffic and increasing complexities. Various intelligent packet transmission strategies have been proposed to tackle different network problems. However, most of the existing research just focuses on the network related improvements and neglects the analysis about the computation consumptions. In this paper, we propose a Value Iteration Architecture based Deep Learning (VIADL) method to conduct routing design to address the limitations of existing deep learning based routing algorithms in dynamic networks. Besides the network performance analysis, we also study the complexity of our proposal as well as the resource consumptions in different deployment manners. Moreover, we adopt the Heterogeneous Computing Platform (HCP) to conduct the training and running of the proposed VIADL since the theoretical analysis demonstrates the significant reduction of the time complexity with the multiple GPUs in HCPs. Furthermore, simulation results demonstrate that compared with the existing deep learning based method, our proposal can guarantee more stable network performance when network topology changes.

The fast increasing chip processing capacities driven by the Moore's Law have encouraged the academia and industry to consider more about general hardware architectures since they allow the repeated use for multiple purposes through the installations of applications. Some techniques utilizing the general hardware architectures have been developed to improve the flexibility of computer networks, such as the Software Defined Networking (SDN) and the Network Functions Virtualization (NFV). For these networks, the applications are required to be computation/communication-efficient since the installed applications share the hardware. In this paper, we study the resource-limited disaster recovery networks constructed by the Movable and Deployable Resource Units (MDRUs) which consist of various general computation platforms. We propose an efficient safety confirmation method through the photo sharing by the survivors. In the proposal, the Absorbing Markov Chain is utilized to model the safety confirmation process, transition matrix of which can be adopted to choose the suitable photo size for optimizing the traffic overhead and buffer consumption. Through periodical update of the photo database, unnecessary packet transmissions can be further avoided with reasonable sacrifice of the computation overhead. To expedite the computation, the GPU-accelerated MDRU is considered to conduct the matrix calculations in a parallel fashion.

The high-speed transmission has become extremely important with the rapid growth of network traffic in wireless networks. Because the available bandwidth of wireless channels are limited, Partially Overlapping Channels (POCs) are widely used in wireless networks to maximize the utilization of channel resources. However, with the traffic patterns of wireless networks becoming huge and dynamic, conventional POC assignment algorithms only designed for constantly generated network traffic are not suitable for the new generation wireless networks. Therefore, in this article, a joint deep Covolutional Neural Network (CNN) and activity vector based intelligent channel assignment algorithm is proposed, which is referred to as CNNAV. With the proposed CNNV approach, the network can learn from the historical traffic patterns and intelligently assign POCs to wireless links. The simulation result shows that, the network performance of our proposal in terms of both packets loss rate and network throughput are better than conventional POC assignment algorithms.

Due to the fast increase of sensing data and quick response requirement in the Internet of Things (IoT) delivery network, the high speed transmission has emerged as an important issue. Assigning suitable channels in the wireless IoT delivery network is a basic guarantee of high speed transmission. However, the high dynamics of traffic load (TL) make the conventional fixed channel assignment algorithm ineffective. Recently, the software defined networking-based IoT (SDN-IoT) is proposed to improve the transmission quality. Besides this, the intelligent technique of deep learning is widely researched in high computational SDN. Hence, we first propose a novel deep learning-based TL prediction algorithm to forecast future TL and congestion in network. Then, a deep learning-based partially channel assignment algorithm is proposed to intelligently allocate channels to each link in the SDN-IoT network. Finally, we consider a deep learning-based prediction and partially overlapping channel assignment to propose a novel intelligent channel assignment algorithm, which can intelligently avoid potential congestion and quickly assign suitable channels in SDN-IoT. The simulation result demonstrates that our proposal significantly outperforms conventional channel assignment algorithms.

Recently, network operators are confronting the challenge of exploding traffic and more complex network environments due to the increasing number of access terminals having various requirements for delay and package loss rate. However, traditional routing methods based on the maximum or minimum single metric value aim at improving the network quality of only one aspect, which makes them become incapable to deal with the increasingly complicated network traffic. Considering the improvement of deep learning techniques in recent years, in this paper, we propose a smart packet routing strategy with Tensor-based Deep Belief Architectures (TDBAs) that considers multiple parameters of network traffic. For better modeling the data in TDBAs, we use the tensors to represent the units in every layer as well as the weights and biases. The proposed TDBAs can be trained to predict the whole paths for every edge router. Simulation results demonstrate that our proposal outperforms the conventional Open Shortest Path First (OSPF) protocol in terms of overall packet loss rate and average delay per hop.

Recent years, Software Defined Routers (SDRs) (programmable routers) have emerged as a viable solution to provide a cost-effective packet processing platform with easy extensibility and programmability. Multi-core platforms significantly promote SDRs' parallel computing capacities, enabling them to adopt artificial intelligent techniques, i.e., deep learning, to manage routing paths. In this paper, we explore new opportunities in packet processing with deep learning to inexpensively shift the computing needs from rule-based route computation to deep learning based route estimation for high-throughput packet processing. Even though deep learning techniques have been extensively exploited in various computing areas, researchers have, to date, not been able to effectively utilize deep learning based route computation for high-speed core networks. We envision a supervised deep learning system to construct the routing tables and show how the proposed method can be integrated with programmable routers using both Central Processing Units (CPUs) and Graphics Processing Units (GPUs). We demonstrate how our uniquely characterized input and output traffic patterns can enhance the route computation of the deep learning based SDRs through both analysis and extensive computer simulations. In particular, the simulation results demonstrate that our proposal outperforms the benchmark method in terms of delay, throughput, and signaling overhead.